GaussMemory：面向长时程机器人操作的任务驱动三维高斯场景记忆
GaussMemory: Task-Driven 3D Gaussian Scene Memory for Long-Horizon Robotic Manipulation
浏览论文内容
中文总结 AI 辅助
该研究针对现有三维机器人记忆系统被动存储的缺陷，提出任务驱动的主动场景记忆框架GaussMemory，在LIBERO、VLABench基准上超越了MemoryVLA、π₀-FAST等方法。
中文摘要 AI 辅助
长时程机器人操作从根本上依赖持久的空间记忆。然而，现有的三维记忆系统仅作为被动记录器运行：它们使用固定的手工规则存储观测结果，对每个场景元素——无论是关键的抓取目标还是无关的背景墙——都赋予同等重要性。在本文中，我们提出从被动存储向主动、任务驱动的空间记忆的范式转变。我们认为，机器人的记忆不应仅记录所见内容，而应主动学习如何记忆——精确发现需跟踪的对象、更新的强度以及需丢弃的内容，所有这些均通过端到端学习实现，无需手工设计规则。关键在于，这种主动范式通过将记忆更新与读出统一为同一认知过程的两个方面来实现，从而实现双向流动：任务需求塑造更新策略，反之亦然。为实现这一愿景，我们引入GaussMemory，它利用三维高斯溅射作为持久几何基底。在LIBERO上，GaussMemory在Goal和Long-10任务上的性能优于MemoryVLA；在VLABench上，它超越π₀-FAST，提升幅度为Track 1任务+5.2%、Track 6任务+6.0%。
英文摘要
Long-horizon robotic manipulation fundamentally relies on persistent spatial memory. However, existing 3D memory systems function merely as passive recorders: they store observations using fixed, hand-crafted rules, treating every scene element--whether a critical grasp target or an irrelevant background wall--with equal importance. In this paper, we propose a paradigm shift from passive storage to active, task-driven spatial memory. We argue that a robot's memory should not simply record what it sees, but actively learn how to remember--discovering which objects to track precisely, how aggressively to update them, and what to discard, all learned end-to-end without hand-designed rules. Crucially, this active paradigm is realized by unifying memory update and readout as two sides of the same cognitive process, enabling bidirectional flow where task needs shape update strategies and vice versa. To instantiate this vision, we introduce GaussMemory, which leverages 3D Gaussian Splatting as a persistent geometric substrate. On LIBERO, GaussMemory outperforms MemoryVLA on Goal and Long-10; on VLABench, it surpasses $π_0$-FAST by +5.2% (Track 1) and +6.0% (Track 6).
发表机构
- Research Institute for Science & Technology, Tokyo University of Science（东京理科大学科学技术研究所）
机构由 AI 辅助整理，请以论文原文为准。
